Re: role = 'img'

Hi Wayne and Wilco,

This is related to:
https://github.com/w3c/wcag21/issues/297

A draft is  in the Wiki:
Providing a Semantically Identified Icon Font with role=img
https://www.w3.org/WAI/GL/wiki/Providing_a_Semantically_Identified_Icon_Font_with_role%3Dimg

GitHub placeholder:
https://github.com/w3c/wcag/blob/tech-icon-font-img-role/techniques/aria/icon-font-img-role.html

Kindest Regards,
Laura

On 10/8/18, Wayne Dick <wayneedick@gmail.com> wrote:
> Hi Wilco,
> I'm thinking of a selector like [role~="img"]. This would search for "img"
> in the role string. It would work with your case. That is about as complex
> as attribute selectors can get. For example we cannot look for
> [style.font-family ~="..."] in CSS. So we need a good top level
> deterministic attribute for selection, role="... img" would work.
>
> Best, Wayne
>
> On Mon, Oct 8, 2018 at 11:24 AM Wilco Fiers <wilco.fiers@deque.com> wrote:
>
>> Hey Wayne,
>> I understand the problem you are outlining, but I'm not entirely clear on
>> what approach you are trying to take? Are you suggesting that you want to
>> use ARIA role=img to annotate font icons, so that if you write a custom
>> stylesheet, you can leave those fonts unchanged? If so, it seems to me
>> there are some problems with adding role=img. With ARIA 1.1 fallback
>> roles,
>> you can have something like role="graphics-symbol img". Are you
>> suggesting fallback roles wouldn't be allowed, or that the best practice
>> is
>> to always add img to the role, regardless of what other roles might be
>> there?
>>
>> Wilco
>>
>>
>> On Mon, Oct 8, 2018 at 7:37 PM Wayne Dick <wayneedick@gmail.com> wrote:
>>
>>> Dear Group,
>>> Right now icon fonts are the only barrier to replacing the author's
>>> font.
>>> The text width problem that was anticipated earlier has not materialized
>>> to
>>> be a problem.
>>>
>>> So, I am hoping this can be a best practice with strong encouragement.
>>>
>>> It really is a total misuse of semantics. People use text element tags
>>> like <i> and <span> to carry pictures.
>>>
>>> This would be innocuous if these were not targets of font change. The
>>> <i>
>>> is important to low vision because it changes the way font looks in a
>>> bad
>>> way. <span> is important because it has no natural semantics.
>>>
>>> If this problem was solved the issue of changing font could be addressed
>>> with font-family style, but without use of role an extension approach is
>>> needed. Then the at can only guess. Does class contain a particular set
>>> of
>>> values? Is the included text one character long?
>>>
>>> A font analyzer has now unambiguous way to separate text for font
>>> replacement and pictures disguised as text to leave alone.
>>>
>>> I need some help, if I am to automate stylesheet selection for
>>> non-experts.
>>>
>>> Best, Wayne
>>>
>>
>>
>> --
>> *Wilco Fiers*
>> Senior Accessibility Engineer - Co-facilitator WCAG-ACT - Chair Auto-WCAG
>>
>


-- 
Laura L. Carlson

Received on Monday, 8 October 2018 19:36:15 UTC